- Some experts think so-called edge computing will be the next big trend in tech after cloud computing.
- In edge computing, a significant amount of processing power is shifted from the cloud to the edge of the network to devices such as manufacturing robots, self-adjusting thermostats, and self-driving cars.
- Helping drive the trend is the growing number of devices that rely on artificial intelligence, which requires lots of data that often has to be processed in real time, making it harder to depend on the cloud.
One of the biggest trends in tech over the last decade has been the shift to cloud computing – the move to store and process data and run applications in data centres maintained companies including Amazon and Microsoft rather than on corporations’ own computers.
But even as that trend is continuing and cloud computing is becoming more important, we’re starting to see the first signs of what might replace or improve upon it – so-called edge or fog computing. This new type of computing has huge potential, boosted not only by the cloud trend but also cutting-edge devices that have the chance to become mass-market products, most notably self-driving cars and robots.
While a lot of work will still be done in cloud data centres, the edge trend will shift much of the critical processing and decision making to devices and gadgets that are connected to them, but at the far reaches of the network.
“There’s going to be a symbiotic relationship between the edge and the cloud,” said Peter Levine, general partner at venture-capital firm Andreessen Horowitz.
Edge computing is shifting processing power out of the cloud and nearer to end users
The term edge computing refers to the concept of a so-called edge device. If you think about networks as a spoke and wheel, with servers and data centres at their centres and things like PCs, smartphones, and tablets arrayed around them in a circle, the latter group of gadgets are edge devices. They’re at the network’s edge.
Your smartphone is attached to the edge of your carrier’s network. If you have a Nest smart thermostat or a Sony PlayStation 4, they’re at the edge of your home Wi-Fi network.
In edge computing, gadgets are still connected to the internet and can still tap into cloud computing services. But they typically have more onboard computing power than in the past and can do more things on their own.
It’s about “moving processing and storage close to the application,” Levine said.
In some ways, with edge computing, the technology industry is going back to the future. Prior to the 1980s, businesses depended on centralised mainframe computers, which workers might interact with using so-called dumb terminals. But PCs shifted processing power out to workers’ desks, out on the edge.
More recently, the cloud-computing trend centralised processing power again. With edge computing, the pendulum is starting to swing back, distributing computing power more broadly, closer to end users.
It’s still too early to get a sense of the potential size of the market for edge computing in dollars. However, the research firm Gartner recently estimated that by 2022, half of all data generated by businesses will be from these smarter edge devices, rather than from the cloud or their own data centres.
The definition of what would be considered an edge computing device is somewhat open to interpretation. But Levine and others think the distinguishing characteristic of such gadgets is their ability to do smart processing on their own. They wouldn’t consider a simple internet-connected lightbulb an edge computer, for example. But they might consider a thermostat that adjusted itself based on your habits to be one.
Artificial intelligence and self-driving cars are driving the edge-computing trend
A big driver of edge computing will likely be the rapid development of artificial intelligence, which can require lots processing power to be available immediately. Some manufacturing robots, security cameras, and augmented-reality headsets already rely on AI. But the technology product that perhaps best highlights the need for and promise of edge computing is the self-driving car.
To be safe on the road, autonomous vehicles, such as those being developed by Waymo, Google’s corporate sibling, have to make sure they’re keeping to their lanes, recognising and stopping at red lights and stop signs, and identifying pedestrians and bicyclists and making sure to yield to them. All of that requires the cars to crunch massive amounts of data in real time, every second of every ride. Indeed self-driving cars generate as much as a gigabyte of data every second, according to some estimates.
It’s so much data that has to be processed so quickly it would be impossible for such cars to rely on servers in the cloud. If they were to depend on Amazon Web Services, say, or Microsoft Azure, to handle such information, they’d have to upload their data to those cloud services, wait for it to be processed, and then wait for the results. Even though those services can be pretty fast, they’re not fast enough to be able to respond in real-time to current driving conditions or immediate dangers.
“The car would blow through the stop signs and run over people before the information got back from the cloud,” Levine said.
So instead of relying on the cloud, Waymo and other self-driving car developers have been loading up their vehicles with heavy-duty processors from the likes of Nvidia and Intel, giving each car enough computing power that it essentially becomes a “data center on wheels,” as Levine puts it.
Edge computing is likely to work in tandem with the cloud, not completely replace it
But the important thing about the new edge-computing trend is that even though it’s pushing processing power back to the edge, it’s not leaving the cloud out. Instead, cloud computing still plays an important role.
With self-driving cars, for example, the vehicles typically, at the end of a day of driving, send the data they have collected to the cloud. The manufacturers of the autonomous vehicle systems use the data they have gleaned from those cars to train and refine their software so the vehicles offer smoother and safer rides.
Robots, phones and web cameras that have built-in facial-recognition systems, and other gadgets that rely on artificial intelligence work in similar ways. The device does much of the initial or immediate processing, but it relies on cloud intelligence to improve how it functions.
When Satya Nadella first took over as Microsoft’s CEO, he famously declared the company would pursue a “cloud-first, mobile-first” strategy. Last year, he updated that vision a little bit. Now, Nadella said, Microsoft is about “intelligent cloud and intelligent edge.”
The software giant has an opportunity to help clients make edge devices work better with the software they’re running on Azure, said Sam George, a director at Microsoft who’s leading much of the company’s edge computing efforts.
Some of Microsoft’s big corporate customers are already benefiting from the edge-computing model, George said. One client is using Microsoft’s edge-computing services with its manufacturing robots. The robots have enough artificial intelligence built-in that they can keep working even when the internet goes out. Without that connection, the company could experience “catastrophic downtime,” he said.
“They’re doing the real-time processing on the device itself,” George said.
Ultimately, when it comes to edge and cloud computing, they’re not necessarily in competition; they don’t have to present users with an either-or proposition, George said. Instead, the two can complement each other, and the important thing for technology providers and their customers is figuring out the right balance between how much processing can be done in the cloud and how much should be done on the edge devices.
“The whole gist of edge is the ability to have consistency between services you run in the cloud and the devices that you run,” George said.